lattifai 0.4.6__py3-none-any.whl → 1.0.0__py3-none-any.whl

lattifai/io/writer.py

@@ -1,102 +0,0 @@
-import json
-from abc import ABCMeta
-from typing import Any, List, Optional
-
-import pysubs2
-from lhotse.supervision import AlignmentItem
-from lhotse.utils import Pathlike
-
-from .reader import Supervision
-
-
-class SubtitleWriter(ABCMeta):
-    """Class for writing subtitle files with optional word-level alignment."""
-
-    @classmethod
-    def write(cls, alignments: List[Supervision], output_path: Pathlike) -> Pathlike:
-        if str(output_path)[-4:].lower() == ".txt":
-            with open(output_path, "w", encoding="utf-8") as f:
-                for sup in alignments:
-                    word_items = parse_alignment_from_supervision(sup)
-                    if word_items:
-                        for item in word_items:
-                            f.write(f"[{item.start:.2f}-{item.end:.2f}] {item.symbol}\n")
-                    else:
-                        text = f"{sup.speaker} {sup.text}" if sup.speaker is not None else sup.text
-                        f.write(f"[{sup.start:.2f}-{sup.end:.2f}] {text}\n")
-
-        elif str(output_path)[-5:].lower() == ".json":
-            with open(output_path, "w", encoding="utf-8") as f:
-                # Enhanced JSON export with word-level alignment
-                json_data = []
-                for sup in alignments:
-                    sup_dict = sup.to_dict()
-                    json_data.append(sup_dict)
-                json.dump(json_data, f, ensure_ascii=False, indent=4)
-        elif str(output_path).lower().endswith(".textgrid"):
-            from tgt import Interval, IntervalTier, TextGrid, write_to_file
-
-            tg = TextGrid()
-            supervisions, words, scores = [], [], {"utterances": [], "words": []}
-            for supervision in sorted(alignments, key=lambda x: x.start):
-                text = (
-                    f"{supervision.speaker} {supervision.text}" if supervision.speaker is not None else supervision.text
-                )
-                supervisions.append(Interval(supervision.start, supervision.end, text or ""))
-                # Extract word-level alignment using helper function
-                word_items = parse_alignment_from_supervision(supervision)
-                if word_items:
-                    for item in word_items:
-                        words.append(Interval(item.start, item.end, item.symbol))
-                        if item.score is not None:
-                            scores["words"].append(Interval(item.start, item.end, f"{item.score:.2f}"))
-                if supervision.has_custom("score"):
-                    scores["utterances"].append(
-                        Interval(supervision.start, supervision.end, f"{supervision.score:.2f}")
-                    )
-
-            tg.add_tier(IntervalTier(name="utterances", objects=supervisions))
-            if words:
-                tg.add_tier(IntervalTier(name="words", objects=words))
-
-            if scores["utterances"]:
-                tg.add_tier(IntervalTier(name="utterance_scores", objects=scores["utterances"]))
-            if scores["words"]:
-                tg.add_tier(IntervalTier(name="word_scores", objects=scores["words"]))
-
-            write_to_file(tg, output_path, format="long")
-        else:
-            subs = pysubs2.SSAFile()
-            for sup in alignments:
-                # Add word-level timing as metadata in the subtitle text
-                word_items = parse_alignment_from_supervision(sup)
-                if word_items:
-                    for word in word_items:
-                        subs.append(
-                            pysubs2.SSAEvent(start=int(word.start * 1000), end=int(word.end * 1000), text=word.symbol)
-                        )
-                else:
-                    text = f"{sup.speaker} {sup.text}" if sup.speaker is not None else sup.text
-                    subs.append(pysubs2.SSAEvent(start=int(sup.start * 1000), end=int(sup.end * 1000), text=text or ""))
-            subs.save(output_path)
-
-        return output_path
-
-
-def parse_alignment_from_supervision(supervision: Any) -> Optional[List[AlignmentItem]]:
-    """
-    Extract word-level alignment items from Supervision object.
-
-    Args:
-        supervision: Supervision object with potential alignment data
-
-    Returns:
-        List of AlignmentItem objects, or None if no alignment data present
-    """
-    if not hasattr(supervision, "alignment") or not supervision.alignment:
-        return None
-
-    if "word" not in supervision.alignment:
-        return None
-
-    return supervision.alignment["word"]

lattifai/tokenizer/__init__.py

@@ -1,3 +0,0 @@
-from .tokenizer import AsyncLatticeTokenizer, LatticeTokenizer
-
-__all__ = ["LatticeTokenizer", "AsyncLatticeTokenizer"]

lattifai/workers/__init__.py

@@ -1,3 +0,0 @@
-from .lattice1_alpha import Lattice1AlphaWorker
-
-__all__ = ["Lattice1AlphaWorker"]

lattifai/workflows/__init__.py

@@ -1,34 +0,0 @@
-"""
-LattifAI Agentic Workflows
-
-This module provides agentic workflow capabilities for automated processing
-of multimedia content through intelligent agent-based pipelines.
-"""
-
-# Import transcript processing functionality
-from lattifai.io import (
-    ALL_SUBTITLE_FORMATS,
-    INPUT_SUBTITLE_FORMATS,
-    OUTPUT_SUBTITLE_FORMATS,
-    SUBTITLE_FORMATS,
-    GeminiReader,
-    GeminiWriter,
-)
-
-from .agents import YouTubeSubtitleAgent
-from .base import WorkflowAgent, WorkflowResult, WorkflowStep
-from .file_manager import FileExistenceManager
-
-__all__ = [
-    "WorkflowAgent",
-    "WorkflowStep",
-    "WorkflowResult",
-    "YouTubeSubtitleAgent",
-    "FileExistenceManager",
-    "GeminiReader",
-    "GeminiWriter",
-    "SUBTITLE_FORMATS",
-    "INPUT_SUBTITLE_FORMATS",
-    "OUTPUT_SUBTITLE_FORMATS",
-    "ALL_SUBTITLE_FORMATS",
-]

lattifai/workflows/agents.py

@@ -1,12 +0,0 @@
-"""
-Subtitle Agents
-
-An agentic workflow for processing YouTube(or more) videos through:
-1. URL processing and audio download
-2. Gemini 2.5 Pro transcription
-3. LattifAI alignment
-"""
-
-from .youtube import YouTubeSubtitleAgent
-
-__all__ = ["YouTubeSubtitleAgent"]

lattifai/workflows/gemini.py

@@ -1,167 +0,0 @@
-"""
-Gemini 2.5 Pro transcription module
-"""
-
-import asyncio
-from typing import Optional
-
-# Import Google GenAI SDK
-from google import genai
-from google.genai.types import GenerateContentConfig, Part, ThinkingConfig
-
-from .base import setup_workflow_logger
-from .prompts import get_prompt_loader
-
-
-class GeminiTranscriber:
-    """Gemini 2.5 Pro audio transcription using the specified Gem
-
-    Configuration (in __init__):
-        - api_key: Gemini API key (required)
-
-    Runtime parameters (in __call__):
-        - youtube_url: YouTube URL to transcribe
-    """
-
-    # The specific Gem URL provided by the user
-    GEM_URL = "https://gemini.google.com/gem/1870ly7xvW2hU_umtv-LedGsjywT0sQiN"
-
-    def __init__(self, api_key: Optional[str] = None):
-        self.api_key = api_key
-        self.logger = setup_workflow_logger("gemini")
-        self.prompt_loader = get_prompt_loader()
-
-        if not self.api_key:
-            self.logger.warning(
-                "⚠️ Gemini API key not provided. API key will be required when calling transcription methods."
-            )
-
-    async def __call__(self, youtube_url: str) -> str:
-        """Main entry point for transcription"""
-        return await self.transcribe_url(youtube_url)
-
-    async def transcribe_url(self, youtube_url: str) -> str:
-        """
-        Transcribe audio from YouTube URL using Gemini 2.5 Pro Gem
-
-        Args:
-            youtube_url: YouTube URL to transcribe
-
-        Returns:
-            Transcribed text
-        """
-        if not self.api_key:
-            raise ValueError("Gemini API key is required for transcription")
-
-        self.logger.info(f"🎤 Starting Gemini transcription for: {youtube_url}")
-
-        try:
-            # Initialize client
-            client = genai.Client(api_key=self.api_key)
-
-            # Load prompt from Gem configuration
-            system_prompt = self.prompt_loader.get_gemini_transcription_prompt()
-
-            # Generate transcription with extended thinking
-            self.logger.info("🔄 Sending request to Gemini 2.5 Pro...")
-            config = GenerateContentConfig(
-                system_instruction=system_prompt,
-                # Enable thinking by including it in response modalities
-                response_modalities=["TEXT"],
-                thinking_config=ThinkingConfig(
-                    include_thoughts=False,
-                    thinking_budget=-1,
-                ),
-            )
-            response = await asyncio.get_event_loop().run_in_executor(
-                None,
-                lambda: client.models.generate_content(
-                    model="gemini-2.5-pro",
-                    contents=Part.from_uri(file_uri=youtube_url, mime_type="video/*"),
-                    config=config,
-                ),
-            )
-
-            if not response.text:
-                raise RuntimeError("Empty response from Gemini API")
-
-            transcript = response.text.strip()
-
-            self.logger.info(f"✅ Transcription completed: {len(transcript)} characters")
-            return transcript
-
-        except ImportError:
-            raise RuntimeError("Google GenAI SDK not installed. Please install with: pip install google-genai")
-        except Exception as e:
-            self.logger.error(f"Gemini transcription failed: {str(e)}")
-            raise RuntimeError(f"Gemini transcription failed: {str(e)}")
-
-    async def transcribe_file(self, media_file_path: str) -> str:
-        """
-        Transcribe audio/video from local file using Gemini 2.5 Pro
-
-        Args:
-            media_file_path: Path to local audio file
-
-        Returns:
-            Transcribed text
-        """
-        if not self.api_key:
-            raise ValueError("Gemini API key is required for transcription")
-
-        self.logger.info(f"🎤 Starting Gemini transcription for file: {media_file_path}")
-
-        try:
-            # Initialize client
-            client = genai.Client(api_key=self.api_key)
-
-            # Load prompt from Gem configuration
-            system_prompt = self.prompt_loader.get_gemini_transcription_prompt()
-
-            # Upload audio file
-            self.logger.info("📤 Uploading audio file to Gemini...")
-            media_file = client.files.upload(path=media_file_path)
-
-            # Generate transcription with extended thinking
-            # Note: For thinking mode, you may want to use 'gemini-2.0-flash-thinking-exp' or similar models
-            self.logger.info("🔄 Sending transcription request...")
-            config = GenerateContentConfig(
-                system_instruction=system_prompt,
-                # Enable thinking by including it in response modalities
-                response_modalities=["TEXT"],
-                thinking_config=ThinkingConfig(
-                    include_thoughts=False,
-                    thinking_budget=-1,
-                ),
-            )
-            response = await asyncio.get_event_loop().run_in_executor(
-                None,
-                lambda: client.models.generate_content(
-                    model="gemini-2.5-pro",
-                    contents=Part.from_uri(file_uri=media_file.uri, mime_type=media_file.mime_type),
-                    config=config,
-                ),
-            )
-
-            if not response.text:
-                raise RuntimeError("Empty response from Gemini API")
-
-            transcript = response.text.strip()
-
-            self.logger.info(f"✅ Transcription completed: {len(transcript)} characters")
-            return transcript
-
-        except ImportError:
-            raise RuntimeError("Google GenAI SDK not installed. Please install with: pip install google-genai")
-        except Exception as e:
-            self.logger.error(f"Gemini transcription failed: {str(e)}")
-            raise RuntimeError(f"Gemini transcription failed: {str(e)}")
-
-    def get_gem_info(self) -> dict:
-        """Get information about the Gem being used"""
-        return {
-            "gem_name": "Audio Transcription Gem",
-            "gem_url": self.GEM_URL,
-            "model": "Gemini 2.5 Pro",
-            "description": "Specialized Gem for media content transcribe",
-        }

lattifai/workflows/prompts/README.md

@@ -1,22 +0,0 @@
-# Workflow Prompts
-
-This directory contains system prompts and instructions for various AI models used in workflows.
-
-## Structure
-
-- `gemini/` - Prompts for Gemini models
-  - `transcription_gem.txt` - System prompt from the Audio Transcription Gem
-
-## Usage
-
-Prompts are loaded automatically by their respective workflow modules. To update a prompt:
-
-1. Edit the corresponding `.txt` file
-2. The changes will be picked up on the next workflow run
-3. No code changes needed
-
-## Adding New Prompts
-
-1. Create a subdirectory for the model/service (e.g., `openai/`, `anthropic/`)
-2. Add prompt files as `.txt` files with descriptive names
-3. Update the loader in the corresponding workflow module

lattifai/workflows/prompts/gemini/README.md

@@ -1,24 +0,0 @@
-# Gemini Workflow Prompts
-
-This directory contains system prompts for Gemini-based workflows.
-
-## Files
-
-### `transcription_gem.txt` [@dotey](https://x.com/dotey/status/1971810075867046131)
-
-System prompt extracted from the Audio/Video Transcription Gem:
-- **Gem URL**: https://gemini.google.com/gem/1870ly7xvW2hU_umtv-LedGsjywT0sQiN
-- **Model**: Gemini 2.5 Pro
-- **Purpose**: Audio/Video transcription with accuracy and natural formatting
-
-## Updating the Prompt
-
-To update the transcription behavior:
-
-1. Edit `transcription_gem.txt` directly
-2. Changes take effect on the next workflow run
-3. No code changes needed
-
-## Gem Information
-
-The original Gem configuration is preserved in the `GeminiTranscriber.GEM_URL` constant and can be accessed via the `get_gem_info()` method.

lattifai/workflows/prompts/gemini/transcription_gem.txt

@@ -1,81 +0,0 @@
-# Role
-You are an expert transcript specialist. Your task is to create a perfectly structured, verbatim transcript of a video.
-
-# Objective
-Produce a single, cohesive output containing the parts in this order:
-1.  A Video Title
-2.  A **Table of Contents (ToC)**
-3.  The **full, chapter-segmented transcript**
-
-* Use the same language as the transcription for the Title and ToC.
-
-# Critical Instructions
-
-## 1. Transcription Fidelity: Verbatim & Untranslated
-* Transcribe every spoken word exactly as you hear it, including filler words (`um`, `uh`, `like`) and stutters.
-* **NEVER translate.** If the audio is in Chinese, transcribe in Chinese. If it mixes languages (e.g., "这个 feature 很酷"), your transcript must replicate that mix exactly.
-
-## 2. Speaker Identification
-* **Priority 1: Use metadata.** Analyze the video's title and description first to identify and match speaker names.
-* **Priority 2: Use audio content.** If names are not in the metadata, listen for introductions or how speakers address each other.
-* **Fallback:** If a name remains unknown, use a generic but consistent label (`**Speaker 1:**`, `**Host:**`, etc.).
-* **Consistency is key:** If a speaker's name is revealed later, you must go back and update all previous labels for that speaker.
-
-## 3. Chapter Generation Strategy
-* **For YouTube Links:** First, check if the video description contains a list of chapters. If so, use that as the primary basis for segmenting the transcript.
-* **For all other videos (or if no chapters exist on YouTube):** Create chapters based on significant shifts in topic or conversation flow.
-
-## 4. Output Structure & Formatting
-
-* **Timestamp Format**
-* All timestamps throughout the entire output MUST use the exact `[HH:MM:SS]` format (e.g., `[00:01:23]`). Milliseconds are forbidden.
-
-* **Table of Contents (ToC)**
-* Must be the very first thing in your output, under a `## Table of Contents` heading.
-* Format for each entry: `* [HH:MM:SS] Chapter Title`
-
-* **Chapters**
-* Start each chapter with a heading in this format: `## [HH:MM:SS] Chapter Title`
-* Use two blank lines to separate the end of one chapter from the heading of the next.
-
-* **Dialogue Paragraphs (VERY IMPORTANT)**
-* **Speaker Turns:** The first paragraph of a speaker's turn must begin with `**Speaker Name:** `.
-* **Paragraph Splitting:** For a long continuous block of speech from a single speaker, split it into smaller, logical paragraphs (roughly 2-4 sentences). Separate these paragraphs with a single blank line. Subsequent consecutive paragraphs from the *same speaker* should NOT repeat the `**Speaker Name:** ` label.
-* **Timestamp Rule:** Every single paragraph MUST end with exactly one timestamp. The timestamp must be placed at the very end of the paragraph's text.
-* ❌ **WRONG:** `**Host:** Welcome back. [00:00:01] Today we have a guest. [00:00:02]`
-* ❌ **WRONG:** `**Jane Doe:** The study is complex. We tracked two groups over five years to see the effects. [00:00:18] And the results were surprising.`
-* ✅ **CORRECT:** `**Host:** Welcome back. Today we have a guest. [00:00:02]`
-* ✅ **CORRECT (for a long monologue):**
-`**Jane Doe:** The study is complex. We tracked two groups over a five-year period to see the long-term effects. [00:00:18]
-
-And the results, well, they were quite surprising to the entire team. [00:00:22]`
-
-* **Non-Speech Audio**
-* Describe significant sounds like `[Laughter]` or `[Music starts]`, each on its own line with its own timestamp: `[Event description] [HH:MM:SS]`
-
----
-### Example of Correct Output
-
-## Table of Contents
-* [00:00:00] Introduction and Welcome
-* [00:00:12] Overview of the New Research
-
-## [00:00:00] Introduction and Welcome
-
-**Host:** Welcome back to the show. Today, we have a, uh, very special guest, Jane Doe. [00:00:01]
-
-**Jane Doe:** Thank you for having me. I'm excited to be here and discuss the findings. [00:00:05]
-
-## [00:00:12] Overview of the New Research
-
-**Host:** So, Jane, before we get into the nitty-gritty, could you, you know, give us a brief overview for our audience? [00:00:14]
-
-**Jane Doe:** Of course. The study focuses on the long-term effects of specific dietary changes. It's a bit complicated but essentially we tracked two large groups over a five-year period. [00:00:21]
-
-The first group followed the new regimen, while the second group, our control, maintained a traditional diet. This allowed us to isolate variables effectively. [00:00:28]
-
-[Laughter] [00:00:29]
-
-**Host:** Fascinating. And what did you find? [00:00:31]
----
-Begin transcription now. Adhere to all rules with absolute precision.